Oxford Handbooks Online: Quantitative Methodology

Quantitative Methodology
Oxford Handbooks Online

Charles H. Franklin
The Oxford Handbook of Political Methodology
Edited by Janet M. Box-Steffensmeier, Henry E. Brady, and David Collier
Print Publication Date: Aug 2008

Subject: Political Science, Political Methodology, Comparative Politics
Online Publication Date: Sep 2009 DOI: 10.1093/oxfordhb/9780199286546.003.0035
Abstract and Keywords
This article reviews the history of the quantitative methodology institutions, Inter-
university Consortium for Political and Social Research (ICPSR), and the American
Political Science Association's Political Methodology Section. It also highlights the role of
organizations and institutions in promoting and structuring the development of
quantitative methodology in political science. The development of summer programs in
quantitative methods is described. There was a market niche for methodology both as a
subfield on its own, and as a direct contributor to improving substance through improved
methods. The existence of the Society for Political Methodology has increased
expectations for graduate training, at least among those who see their careers as
methodologists.
Keywords: quantitative methodology, summer programs, institutions, political methodology, political science
1 Introduction
THE development of political methodology as a field has been closely associated with the
development of organizations and institutions that support the field. This chapter focuses
on the two organizations most closely associated with quantitative political methodology,
the Inter‐university Consortium for Political and Social Research (ICPSR) Summer
Program and the Society for Political Methodology, which is effectively synonymous with
the American Political Science Association (APSA) organized section on Political
Methodology.1 The approach in this chapter emphasizes the role of organizations and
institutions in promoting and structuring the development of quantitative methodology in
political science. Excellent reviews of the intellectual developments in methodology are
Page 1 of 21
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). © Oxford University Press, 2018. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).
Subscriber: Vanderbilt University Library; date: 17 January 2019

available in Achen (1981); King (1990); and Bartels and Brady (1993). The many chapters
of this Handbook are themselves a profound statement of the intellectual issues in
contemporary methodology. The institutions I describe developed out of intellectual
pressures and needs, which I briefly highlight, but I am more concerned here with
organizations and their role in promoting political methodology.
For historical and intellectual reasons (both good and bad) organizational
(p. 797)
developments in quantitative methodology proceeded largely apart from developments in

qualitative methodology, especially prior to the 1990s which is my focus. The chapter in
this volume by Colin Elman reviews those developments and should be seen as the other
side of the coin described here. Recent scholarship has seen productive and positive
efforts at uniting these different strands of research methodology (King, Keohane, and
Verba 1994; Brady and Collier 2004), even if such a union is not visible in this chapter.
Academic institutions develop and are sustained because there are intellectual and
professional needs that they serve. The development of “summer programs” in
quantitative methods in the 1960s was a response to the needs of a newly quantitative
field that lacked a historical foundation of training in statistical techniques. However, as
we shall see, these programs persist and thrive because contemporary students remain in
need of statistical training outside their home institutions. The development of the
Society for Political Methodology and its subsequent twin, the APSA Political
Methodology Organized Section, arose from a different intellectual pressure—the need
for legitimacy of a subfield and a setting in which methodologists could speak to their
peers in the technical language of a field.
The surge in quantitative analysis in political science in the 1960s grew out of the
behavioral revolution which required new types of evidence for political research. The
skills to carry out this type of research, and the financial resources to conduct national‐
scale survey research, led to pressures for both access to training and the sharing of data
resources. These needs were both met in the development of the Inter‐university
Consortium for Political Research (ICPR, subsequently renamed the Inter‐university
Consortium for Political and Social Research (ICPSR)). ICPR made routinely available to
political scientists data that had previously required special access, often at the cost of a
semester of leave in Ann Arbor. Statistical training that was an integral part of ICPR, the
“Summer Program,” opened the covers of statistics textbooks for both a new generation
of graduate students and for faculty who had received Ph.D.s in an era that treated
statistics as a foreign language (literally). Intellectual and professional pressures made
acquisition of quantitative skills valuable in the 1960s and later, and this pressure
supported the creation and growth of the ICPR Summer Program. Parallel developments
in Europe likewise supported and sustained the foundation of the Essex Summer School
in Social Science Data Analysis and Collection, founded in 1967, four years after the first
Ann Arbor Summer Program.
Page 2 of 21

The pressures that brought about the Society for Political Methodology (SPM) and the
APSA organized section were quite different, coming from an already established group
of political scientists who felt their work was underappreciated and poorly supported by
the professional organizations and conferences of the day. If ICPR and the Ann Arbor and
Essex Summer Programs were primarily about skill acquisition, the SPM was about
extending the research horizons. SPM set as its goal the development of a political
methodology devoted to questions arising from problems unique to political data rather
than borrowing solutions from other disciplines whose (p. 798) concerns might only
tangentially reflect the concerns of political scientists. SPM was also about extending the
boundaries of political methodology by providing settings in which one's research could
be valued for its methodological contributions alone, and establishing the full legitimacy
of methodology as a field within political science. My review of these developments will
focus on intellectual and organizational forces and will deliberately avoid giving much
attention to the efforts of individuals in creating these organizations. A proper history
should acknowledge and celebrate the hard work of the many individuals who helped
create, nurture, and energize these organizations. Nonetheless my focus here is not on
individual contributions but on the broader forces at work.
Page 3 of 21

2 The Development of Summer Programs in

Quantitative Methods
The “behavioral revolution” in political science, both in the United States and in Europe,
produced a sudden rise in demand for technical analysis skills. A discipline which had
only sporadically employed quantitative methods (though with some outstanding
exceptions: Rice 1928, for example, and most impressively Gosnell 1927) suddenly found
that its most impressive work was being produced by scholars comfortable with Hollerith
cards, counter sorters, and even IBM computers. The election studies conducted at the
University of Michigan by Angus Campbell and his junior colleagues Philip Converse,
Warren Miller, and Donald Stokes culminated in The American Voter (Campbell et al.
1960). That book more than any other changed the course of research on American
elections.
The American Voter, it must be stressed, was not primarily innovative in its use of “new”
analytic methodologies. The cross‐tabs and percentages that were the core analytic
method were certainly not new, and the survey designs were considerably simpler than
the multiwave panel study of Erie County, Ohio in the 1940 election conducted by
Lazarsfeld, Bereleson, and Gaudet (1944). It was, rather, the psychological theory
coupled with the determined empirical testing that set The American Voter apart from its
predecessors. The psychological theory was exploited over and over throughout the book,
from partisanship to impressions of the parties and candidates to the effects of cross‐
pressures to the crucial role of psychological attachment to groups, rather than mere
sociological membership in the groups, that set a new course for electoral studies. And
each theoretical assertion was confronted with repeated tests within the data, each
probing the evidence in support of the claim. This was not the first exploitation of survey
analysis, let alone the first quantitative work, in political science. But the behavioral
revolution of the 1950s both supported and was fed by the evident success of the theory
and analytic methods of The American Voter.
A number of quantitative studies had certainly appeared well before The American Voter.
The Columbia University studies of the 1940 and 1944 campaigns set the stage (p. 799)
for later national surveys but the core methodology and analytic techniques were fully
present in those early surveys. Likewise the Michigan Survey Research Center conducted
election studies in 1948, resulting in a small book, The People Elect a President (Campbell
and Kahn 1952), and the 1952 presidential election produced The Voter Decides
(Campbell, Gurin, and Miller 1954), a more substantial development of what was to come.
Yet neither of these studies revolutionized the study of elections in the larger profession.
Likewise the earlier empirical studies by V. O. Key that were part of his Southern Politics
(Key 1949) had contributed to the development of the behavioral revolution, as did Dahl's
subsequent study of New Haven politics in Who Governs? (Dahl 1961). It is appropriate
then to place The American Voter within the pantheon of books that forced a new
Page 4 of 21

approach to political science, and electoral behavior, without giving it an exaggerated

importance.
In the development of quantitative training programs, however, Michigan and the

“Michigan election studies” do hold a special place. This place is due to the coincidence
of two forces: a monopoly on highly valuable data and an entrepreneurial willingness to
give up that monopoly.
By the time The American Voter was published, the Michigan Political Behavior Program
had amassed the only academic collection of national survey data on US elections then
extant. Commercial polls, such as those conducted by the Gallup organization, were not
available for scholarly analysis at that time. This was an incredibly valuable commodity
within political science. Under the existing norms of the time, the “Michigan election
studies” could easily have remained just that— proprietary to Campbell and his co‐
workers. Yet such was not the case, at least for an academic elite. In the 1959–60
academic year, V. O. Key was given full access to the data collected between 1948 and
1958, as well as office space and research assistance in Ann Arbor to conduct his own
analysis of these data (even as The American Voter was still in final manuscript
preparation).2 The resulting Public Opinion and American Democracy was a major
contribution in its own right, but perhaps more importantly it was an exemplar of the
power of secondary analysis if access to data were possible.
At the same time it was apparent that standards for graduate training in political science
had ill equipped new scholars to exploit the data central to the behavioral revolution. V.
O. Key's 1954 A Primer of Statistics for Political Scientists explains in its preface and
demonstrates by its existence the need for even the most elementary instruction in
statistics for political scientists. In addition to a need for introductions to the field of
statistics, political scientists also needed to acquire some more specialized skills in data
processing, then primarily conducted with punch card counter sorters. The techniques
required to produce a two‐way table were at the time considerably more complex than
introducing students to the “cross‐tabs” command is today. By the latter part of the 1950s
new skills were just beginning to be demanded, as illustrated once more by The American
Voter. In 1959 it required special access to an IBM electronic computer, as opposed to
counter sorters, to produce the multiple (p. 800) linear regression that underlies the
analysis of the “six‐component” model of vote choice illustrated by figure 4–2 and
described in footnote 6 on page 73 of Campbell et al. (1960). The skills required for this
were far in advance of any generally available in political science.
The need for new training opportunities was evident in the 1950s. The Social Science
Research Council funded early summer training programs at the SRC in Ann Arbor in
1954 and 1958, introducing students to both survey methodology and techniques of
analysis. Similarly Yale University developed a fellowship program allowing advanced
graduate students to spend time at the SRC. Such opportunities, however, remained
limited and largely unavailable outside of elite academic circles.
Page 5 of 21

Meanwhile, requests for access to the SRC data increased and placed a burden on the
Political Behavior Program to meet the demands of reproducing card decks and shipping
the boxes of cards, as well as raising the question of who should be allowed access to the
data.
By 1962 the accumulated issues of a monopoly on data many political scientists wished to
analyze and the need to find financial resources to support the election studies led to the
creation of the Inter‐university Consortium for Political Research (ICPR). A 1989 report
on the then renamed ICPSR neatly summarizes the motivations for creating this
institution:
As in all successful collaborations there was to be a two‐way flow of benefits

within the new consortium. Continually challenged to raise the funds necessary to
keep the sequence of electoral surveys in flow, Warren E. Miller and his colleagues
of the Political Behavior Program sought in ICPR a device that would generate
some funds directly (the initial membership contribution was set at $2500
annually) and more importantly, allow mobilization of the research community in
support of major grant applications. Opposite numbers in social science
departments, mostly political scientists initially, saw the consortium as a means of
acquiring access for themselves and their students to survey data that were too
expensive to generate locally and as a means also of developing the analytical and
methodological skills necessary for advancement in a social science world that
was changing much faster than the formal curriculum. This latter aspect of the
consortium plan was to involve the summer training of junior faculty and graduate
students and provision of considerable remedial instruction for established faculty
as well. (Blalock 1989)
This captures the entrepreneurial aspects of the early ICPR as well as the collegial
willingness to give away what many at the time would have thought of as a private data
monopoly. Miller and his colleagues found that by converting a private resource into a
public good, they could gain more support for the future enterprise. And political
scientists gained access to data they could never collect on their own. Initial membership
in ICPR remained limited to the handful of research universities that could both afford
the fee and had the research focus that justified such an expense. The charter
membership consisted of just eighteen universities.
Page 6 of 21

The second core

component of the early
ICPR was a summer
training program designed
to acquaint students with
the design and analysis of
surveys and political data
more generally. Here the
ICPR Summer Program
Fig. 35.1. Attendance at ICPSR Summer Program
classes, 1963–2006
was born at the same time
as the data dissimilation
role of the Consortium.
While quantitative training may (p. 801) have advanced somewhat beyond the standards
Key had addressed in his 1954 book, there was still a compelling need for training still
unavailable within political science departments. If such training was unavailable locally
(and implicitly political scientists were unwilling to enroll in classes in statistics
departments, few of which were well versed in survey analysis in any case) then Ann
Arbor in June and July offered a solution.
The first summer program, held in 1963, drew a substantial eighty‐two students and 1964
saw ninety‐one attend. But this more than doubled in 1965 to 229 and around 200 or
more became the norm through the mid‐1970s. The earliest years featured courses in
research design and in quantitative analysis, but the scope of the offerings rapidly
expanded. By the early 1970s the summer offerings covered a range from introductory
statistics through regression to specialized topics such as time series and factor analysis.
By the mid‐1970s offerings included classes in causal models, exploratory data analysis,
and multidimensional scaling. In barely ten years the universe of statistical techniques
made available to political scientists had exploded.
This growth in summer courses reflected the tremendous growth in applied statistical
methods in the research literature. From simple descriptive statistics, crosstabs, and a
rare regression in 1960, by 1975 the profession had discovered and largely borrowed
methods such as factor analysis from psychology, time series from statistics, and
regression and simultaneous equation models from economics. For more on the explosion
on imported techniques during this time see King (1990).
This growth in the breadth of techniques used in the profession seemingly fueled further
growth in the Summer Program offerings and attendance. Figure 35.1 illustrates the
growth of attendance by a factor of ten between 1963 and 2007. This growth is in part
due to good marketing of courses by ICPSR, but also reflects a strong demand which
appears unabated even in recent years.
Page 7 of 21

(p. 802)
If the impetus for the

Summer Program was the
immediate needs of
political scientists who
lacked newly useful
technical skills, we might
have expected enrollments
Fig. 35.2. Faculty and Ph.D. participation in ICPSR to decline over time as
Summer Program classes, 1963–2006
graduate programs
increasingly included
quantitative courses in their local curricula. Certainly few graduate programs included
statistics courses in 1960 and a large majority do so now. Few programs now fail to
include at least introductions to statistics and multiple regression and most include more
advanced courses as well.
Yet there is no evidence in Figure 35.1 that demand for Summer Program classes has
abated over time. This implies that whatever growth in quantitative methodology classes
has occurred within departments over time has not kept up with the aggregate demand
for quantitative training. Of course the changing mix of departments and fields
represented in the Summer Program makes any direct inference about political science
programs suspect. Likewise, growth of enrollment in graduate programs may well have
outstripped Summer Program enrollment growth, so smaller proportions of all graduate
students in political science and other social sciences are attending the Summer
Program. Nonetheless, it is clear that the social sciences send a steadily increasing
number of students to the Summer Program despite the growth of local course options
across all the social sciences.
Likewise there has been some change in the mix of participants in the Summer Program,
yet not wholesale changes. The early Summer Program explicitly stated the need to offer
training to both junior and more senior faculty to help them catch up with the new
demands of the field. Yet if graduate training has allowed much greater opportunities for
quantitative training over the past forty years, we might expect that fewer and fewer
faculty would now require such “remedial” training.
Page 8 of 21

Figure 35.2 shows the

percentage of faculty or
holders of Ph.D.s within
the Summer Program from
1963 through 2006 (the
last year for which such
data are currently (p. 803)
available). There was a
steady rise in faculty
participation through the
mid‐1980s, suggesting that
many faculty did, in that
period, find their graduate
training insufficient for
their professional needs.
Since 1985 the rate of
Fig. 35.3. Distribution of participants by discipline in
ICPSR Summer Program classes, 2007 faculty participation has
declined but has remained
3
close to 25 percent since 1995. If graduate training were providing all the quantitative
skills faculty require, then after some forty‐four years we might have expected this
proportion to have declined.
The mix of disciplines represented in the Summer Program has changed substantially
over the years, from almost exclusively political science in the first years to a much more
heterogeneous mix today. Figure 35.3 shows that political science remains the largest
single discipline represented but stands at just under 30 percent of all participants.4
(p. 804) 2.1 Dynamics of Course Offerings
The earliest days of the Summer Program focused on research design, primarily with
surveys, and a single class in quantitative analysis. That rapidly changed as the number
of participants increased and the breadth of techniques expanded. By the early 1970s
classes included not only introductory statistics but regression and factor analysis.
Lectures on mathematics for social scientists (with a focus on matrix algebra and
calculus) were also added in the 1970s.
As the structure of the program expanded (exploded) beyond the original needs of novice
survey analysts, the structure of the program became clear. The program divided into
what came to be called “tracks”. Track I classes included introductory classes in
mathematics, introductory statistics, and a first class in regression (without matrix
algebra). Ancillary lectures also provided an introduction to statistical computing
(originally required for those having to use the Michigan Terminal System mainframe
operating system and its MIDAS statistical program for the first time and subsequently
Page 9 of 21

broadened to cover the range of popular statistical programs such as SAS, SPSS, and
Stata in later years).
Track II classes included linear models (with matrix algebra) and a variety of courses
focused on such topics as time series, factor analysis, multidimensional scaling,
simultaneous equation systems, latent variable models, and more.
Course offerings followed developments in the broad area of social science statistical
methods. In the 1970s, Exploratory Data Analysis appeared on the curriculum, following
publication of Tukey's book of that name, and stayed on the curriculum 1974–83. Time
series, based on Box— Jenkins models, entered the curriculum around 1976.
The early Summer Program emphasis on survey design continued to have a presence in
the curriculum. In the 1980s and 1990s courses included The Logic of Data Analysis:
Measurement and Design (1982–92), Implications of Sampling Design (1984–8), and
Population Estimation (1987–9).
Beginning in the 1980s and 1990s a variety of classes were added (and later dropped)
which focused on applications of quantitative techniques to special substantive topics.
These classes were in part a response to new directions of topical research and partly
driven by the availability of grants to support the courses. Examples include Asian
American Research Methods (1980–5), Empirical Research and Gender Issues (1982–3),
Latino Research Issues (1987–94), Quantitative Methods of Program Evaluation (1991–2),
Quantitative Analysis and the Study of Africa (1996–2001), Quantitative Approaches to
the Study of Latin America (1993–7), and Quantitative Research on Race and Ethnicity
(since 2006).
Despite the considerable increase in statistical offerings, criticism from the new Society
for Political Methodology led to some changes in course offerings. By the late 1980s the
curriculum had settled into a pattern of primarily linear models, at various levels, along
with dimensional and time‐series classes. Around 1986 political scientists began to push
for new courses that would cover limited and discrete dependent variables and other
topics in which developments were accelerating in the 1980s. (p. 805) Within a few years
the Summer Program added a course in Maximum Likelihood Methods (1990),
Hierarchical Models, Categorical Regression, Latent Growth Models, and Non‐linear
Systems. This refreshing of the curriculum primarily in the 1990s substantially increased
the range of offerings while maintaining the place for introductory classes.
Around 2000, an explicit “Track III” was developed. This set of courses was intended to
push summer participants to embrace the more challenging topics of recent
methodological developments. New classes in this track included Bayesian Methods,
Modern Regression, and Advanced Maximum Likelihood along with existing classes on
Maximum Likelihood, Categorical Analysis, Lisrel Models, and Game Theory.One of the
marks of this new track was an emphasis on computing using the S language and the R
Page 10 of 21

statistical program. A two‐week lecture course on the S language became an integral part
of this effort.
Throughout its history, the Summer Program has maintained a quite stable core of
classes, with a dynamic range of additional offerings. The basic introductions to statistics,
regression, and computing have remained part of the curriculum since the early 1970s.
More advanced or specialized courses have also been offered for most of the program's
history, such as time series, dimensional analysis, and structural equation models. And a
number of more recent additions have remained on the curriculum for over a decade,
such as Maximum Likelihood, Game Theory, Advanced Maximum Likelihood, Hierarchical
Linear Models, Categorical Models, and Nonlinear Systems. Still other newer classes
appear set to remain for the foreseeable future, such as Bayesian Methods and Modern
Regression.
Despite that stability, there has also been a highly dynamic set of courses that have come
and gone, some for lack of audience and others due to changing interests and funding
support. The workshops on quantitative analysis of substantive topics are one example.
But statistical offerings have also come and gone: Advanced Analysis of Variance, Event
History Analysis, Item Response Theory, Mathematical Statistics (a disappointed effort to
add a “high end” probability and inference course to the curriculum in the mid‐1990s),
Mixed Models, Pooled Time Series Analysis, Resampling Methods, Spatial Analysis and
Geographic Information Systems, and Vector Autoregression.
2.2 Why do we Continue to Need Summer Programs?
It is said that the early founders of the ICPSR Summer Program expected it to wither
away within a few years.5 The initial assumption may have been that once a core of
political scientists were trained in quantitative methods, they would return to their home
universities and teach subsequent generations, making a summer program unnecessary.
Whether this was ever really the assumption, the past forty years have (p. 806)
demonstrated the growth in demand for summer program instruction, rather than its
demise.
Whether true or not, the anecdote raises the question of why summer programs do
persist, not only at Ann Arbor but at Essex and in newer short‐term programs at Oxford
and in programs sponsored by the ECPR.
Perhaps most puzzling is the continued core of introductory courses that remain part of
the ICPSR Summer Program. Certainly by now virtually all graduate programs offer at
least introductory statistics as part of their curriculum, seeming to obviate the need for
such coursework in the summer. In 2007, for example, approximately 85– 120 Summer
Program students took one of the introductory statistics or introductory regression
classes. These students could have come from any discipline, so they may or may not
represent a large number of political scientists, but the simple fact that at least an eighth
Page 11 of 21

of all 2007 participants chose to take introductory classes shows that there remains an
important role for these classes in the program.
There are a number of possible reasons for this persistence of demand. Some students
may prefer to focus on substantive courses during the academic year, and concentrate
their methodological training during the summer. For these students, the availability of
“local” statistical training is not attractive. Another possibility is that these students are
disproportionately lacking in mathematical background and so benefit from taking
summer courses as either preparation for local courses, or as a means of reviewing
coursework they may have taken but not mastered at their home institution.
This last point is strongly supported by overwhelming evidence that many Summer
Program participants feel the need for more mathematical training. Over 200 students (a
quarter of all participants) attend lectures in either basic mathematics or matrix algebra
and calculus. Whether a refresher or a first exposure, this demand makes clear the
prevalence of weak mathematical training among social science graduate students. Few
undergraduate programs in political science (and most other social sciences outside of
economics) require any mathematical training as part of their degree requirements or as
a requirement for entry into their graduate programs. Likewise, while many
undergraduate programs include a “research methods” class that may include some data
analysis, few require a rigorous sequence in statistics. The consequence of this is a
divergence between the requirements of an undergraduate degree and the skills required
for graduate training.
While only a small minority of undergraduates go on to graduate school in the social

sciences, it appears that very few departments provide even a “pre‐graduate school”
track of suggested courses that would provide their best undergraduates with foundation
courses in mathematics or statistics. The result is that the vast majority of social science
graduate students require some degree of “remedial” mathematics and statistics training,
in the sense that material that could have been covered as an undergraduate must be
learned as a graduate student. This lack of training is reflected in the growth of
mathematics “bootcamps” in a number of political science programs, and continued
demand in the Summer Program.
If something like a quarter of Summer Program students require introductory classes,

about three‐quarters are prepared for more advanced material. This includes (p. 807)
large numbers who take classes in linear regression using matrix algebra. Here students
have often had an introduction to regression at their home institution, but frequently one
that does not include the more general form allowed by matrix algebra. Since linear
models are a fundamental foundation for advanced work, this is a crucial gateway that
the Summer Program provides, and for which large demand remains beyond what is
provided by home institutions.
Page 12 of 21

Still, over half of Summer Program participants take part in classes devoted to statistical
topics beyond that of linear regression. Here the demand seems best explained by the
enormous range of statistical methods and models that currently exist and that are
emerging. And this is perhaps the explanation for why summer programs remain both
popular and necessary beyond the need for introductory classes. The range of statistical
models in the social sciences is huge. For students whose primary focus is not
methodology (and even for some who are methods specialists) there is a great need for
courses in advanced topics that serve their substantive needs.
The 2007 ICPSR Summer Program offered twenty‐seven courses beyond the level of
linear regression in matrix form. Clearly no graduate student has the time to take
academic year courses in twenty‐seven topics, even in an extended graduate career. Yet
each of these courses exists because they provide techniques for addressing a wide range
of substantive topics. Likewise, no social science department has the methodological
focus to offer more than a handful of these topics in a normal curriculum. By pooling
students across the world, summer programs at ICPSR, Essex, and elsewhere provide the
mechanism to allow economically feasible instruction in such specialized topics.
For well‐prepared students, summer programs offer a relatively quick introduction to

advanced topics. In from one to four weeks, students can receive 30–40 hours of
instruction, roughly the equivalent of (or even more than) a semester‐long course. There
are both advantages and disadvantages to this type of course. The daily meetings can
enhance instruction because of the continuity from day to day that allows themes to
remain apparent that might be lost if stretched over once‐a‐week classes during the
semester. The disadvantage of daily classes is that there is little time for reading and
reflection. There is little time to go back and review any points that are missed initially.
One negative aspect of summer courses, but a potential attraction for students, is that the
grading may be less demanding than during an academic semester. Given the limits of
time, exercises are somewhat less numerous and exams are rare. Likewise there is no
time for students to prepare “term” papers during the summer courses. While the quality
of the lectures may be high, the opportunities for students to apply the methods are more
limited than during an academic semester.
Finally, demand for summer programs may in part benefit from the social nature of
summer instruction. Students and faculty alike assemble for the special purpose of taking
and teaching quantitative methodology. (The same applies to programs in qualitative
methods. See the chapter by Elman, this volume.) Attending the program requires a
commitment and some sacrifice by students, and they have this in common with all
participants. The camaraderie that forms among Summer Program (p. 808) participants
almost certainly enhances the experience and promotes collaborative learning of the
material.
Page 13 of 21

2.3 Serving the Needs of Students and the Needs of the Profession
Far from withering away, summer programs in quantitative methodology have only grown
with time. They serve vastly increased numbers of students and offer a wide range of
courses, from elementary to advanced, and from standard to cutting‐edge methodologies.
Some degree of “remedial” instruction in summer programs is likely to remain. This

partially reflects the diversity of graduate programs and graduate students, some of
whom are likely to continue to require opportunities to learn (or relearn) the basics. This
need is not limited to weak programs or weak students. So long as mathematical and
statistical training is not required by undergraduate programs in the social sciences, it
will inevitably be required as part of graduate training. One might question the
desirability of this state of affairs, but despite the “behavioral revolution” of sixty years
ago, the nature of undergraduate training in political science (and other social sciences)
has not mirrored the changes in graduate education. This means that many graduate
students require a venue to acquire basic skills, and summer instruction remains an
attractive option.
At the same time, summer programs have made great strides in providing a wide range of
courses in advanced and state of the art topics. The ICPSR Summer Program, the Essex
Summer School, and others offer topics rarely found on departmental course lists, simply
because the requisite skills (both for instructors and for students) are limited and can
only be successfully marshalled by pooling across many institutions.
As new techniques are constantly developed in quantitative methods, there will also
remain demand for instruction in these new methods. Flexible summer programs are well
suited to providing these because they lack permanent faculty and so can add (or drop)
classes and instructors with relative ease.
3 The Society for Political Methodology and

APSA Organized Section on Political
Methodology
Methodology as a distinct subfield in political science developed slowly in the wake of the
behavioral revolution. While growing numbers of political scientists in the 1960s and
1970s used increasingly sophisticated models in their substantive work, relatively
(p. 809) few considered themselves primarily “methodologists.” Likewise while
departments required the services of faculty who could teach quantitative methods, few
judged tenure cases primarily on methodological contributions, but rather required the
normal substantive contributions to scholarship along with methodological teaching.
Page 14 of 21

The American Journal of Political Science early on provided pages for its methodological
“Workshop,” but there were few other opportunities to publish purely methodological
contributions. The “Workshop” section of AJPS consisted primarily of expository
introductions to techniques, rather than original developments of new statistical
approaches. This reflected both the limitations of the methodology field and the interests
of the profession in quick introductions to particular techniques.
In a run of eleven volumes from 1974 through 1985, Political Methodology provided a
journal outlet for methods research. While the journal had a limited circulation and
focused more on applications of existing methods than on the development of new
techniques, it was the first journal specifically devoted to quantitative methods in political
science.
The 1980s saw changes in political methodology as a field. Stimulated in part by a 1977
textbook, Hanushek and Jackson's Statistical Methods for Social Scientists, the first
serious econometrics textbook aimed at political scientists, more political scientists began
to see their role as methodologists as central to their academic identity, rather than
simply as a useful skill for teaching and research.
In 1981, Achen's “Towards Theories of Data” appeared (Finifter 1981). This chapter
focused on the limitations of political methodology as a field, and focused attention on
two unsolved challenges to the discipline, measurement error and the aggregation
problem. Achen's argument struck a chord with methodologists in the field, and set the
stage for a 1983 Achen paper on ecological inference. The 1983 paper became an
example of how political methodology could address fundamental problems of inference
in political science that were unlikely to be solved by statisticians outside the field. This
in turn illustrated why political science needed a group of scholars committed to
fundamental research on methodological issues. In the discussion following the
presentation of Achen's paper, sitting on the steps in the Palmer House, a group of
methodologists decided to hold a small conference just for methodological issues in the
summer of 1984. This was the origin of what has become the “Summer Methods Meeting”
and more crucially the development of a self‐conscious and activist subfield of political
methodology within APSA.
While it is neat to tie the development of the field to the accidents of Achen's presentation
in 1983, the developments reflect broader issues and discontents within the small
methodology community of the time. If Achen had laid out an intellectual rationale for a
subfield specifically devoted to developing new methodologies, a more prosaic need
among methodologists was increased outlets for methodological work. Professional
conferences provided few if any panels devoted exclusively to methodological papers. The
result was that methodology papers competed with substantive ones, or had to appear as
disguised topical papers, in order to find space on conference programs. As a result,
methodological critique was often limited and purely methodological papers received
scant attention.
Page 15 of 21

At the same time, publication outlets remained limited. The AJPS “Workshop”
(p. 810)
typically printed no more than four articles a year, and no other mainstream journal
devoted space specifically for methodological work. The niche market in Political
Methodology suffered from low visibility, including a lack of indexing in the Social Science
Citation Index. Even that journal was in its death throes in the early 1980s, ceasing
publication in 1985.
A second pressure for a new subfield was that a second generation of methodolo‐ gists
was coming of age. While some of the leaders of the new methodology subfield had
attended graduate school in the 1960s, many had Ph.D.s from the 1970s and a few in the
1980s. These were generally not students for whom counter sorters had been a major
advance. Rather, they had learned statistical methods from econometrics textbooks by
Goldberger or Johnston. Quite a few were also not graduates of undergraduate political
science departments but had instead majored in physics or economics or psychology. This
background left them somewhat better prepared for statistical studies than the typical
undergraduate major in political science. And a few had “grown up” in the new
quantitative political science and had absorbed the need for technical training either as
undergraduates or as graduate students.
The development of the methodology subfield also reflected the competitive pressures of
the post‐behavioral revolution era. Standard statistical techniques were routinely applied
to political data. Yet the standard techniques often left much to be desired. The
ubiquitous dichotomous dependent variables of survey research meant that linear
regression was not the best technique available, though it continued to be widely used.
New sensitivity to the role of measurement error, and the important consequences it had
for substantive conclusions, meant that applied work needed to be more concerned for
arcane methodological issues. Simultaneous equations models of political behavior raised
issues of identification and specification decisions. All of these areas provided
opportunities for methodology both to provide critiques of existing research and to
advance improved approaches to inference. Thus there was a market niche for
methodology both as a subfield on its own, and as a direct contributor to improving
substance through improved methods.
Academic niches can be filled by organized entities, and this was the case with political
methodology. Three related developments became the institutional foundations of the
new political methodology subfield in 1984 and later.
First, an informal group created a small conference in 1984 that became the first
“Summer Methodology Meeting.” Attended by less than twenty people, the first meeting
provided the forum for methodological research that had been lacking within the existing
national and regional professional associations. With a conference devoted exclusively to
methodological research, and with sessions conducted entirely by plenary session, with
intense discussion, the new conference provided both the opportunity for presentation
and the trenchant critique that had been lacking. With a receptive audience, and with
blistering criticism to promote hard and careful work, the Summer Methods Meetings
Page 16 of 21

provided the opportunity for recognition and the incentive to produce new
methodological research. The organizers of this Summer Methodology Meeting became
the Society for Political Methodology.
The Summer Methods Meeting also soon developed a training mission. Starting
(p. 811)
with the fourth Summer Meeting, graduate students were included in the conferences,
with grants providing support for their participation. This self‐conscious attempt to
recruit the next generation of methodologists, and to expose them at an early enough
point in their careers that they could properly anticipate the training needed for a career
in methodology, distinguished the new methodology group as an entity intent on
developing its next generation.
From less than 20 participants at the first Summer Meeting, conference attendance has
grown to over 150 in recent years. This mix of faculty and graduate students provides a
crucial forum for socialization of graduate students to the subfield and for intellectual
stimulation from peers.
The second institutional development was the founding of a new journal devoted to
political methodology. It was agreed that the major journals of the profession were
unlikely to provide sufficient space for purely methodological work, and that in any case a
journal that focused exclusively on methodology would be able to set a higher standard as
well as provide a necessary outlet. This new journal, Political Analysis, first appeared as
an annual volume under the editorship of James Stimson in 1989. The annual nature
reflected an initial concern that the volume of high‐quality articles might be limited, as
well as lack of interest among publishers.
Stimson's introduction to the first volume of Political Analysis sets out the mission
statement of both the journal and the subfield: “Political Analysis aspires to be the
publication of record for methodological developments in political science” (Stimson
1989, x). Seven annual volumes followed, publishing about eight articles and a bit over
200 pages in a typical volume. In 1999, Political Analysis became a quarterly journal and
the volume of articles expanded considerably. Since 2000 the journal has typically
published over twenty‐five articles and over 400 pages each year. This development has
meant that political methodologists have an established outlet for their work, one which
is now recognized as an important journal in the field.
The third institutional development was the creation of an organized section for political
methodology within the American Political Science Association. Organized sections were
introduced within APSA in 1983, and political methodology became the tenth organized
section in 1985. As an organized section, political methodology was able to lay claim to
panels at the annual meeting of APSA, and was able to eventually bundle subscriptions to
Political Analysis with membership in the section. For a new field, political methodology
experienced very rapid growth, becoming the second‐ largest organized section based on
membership. This organizational role within APSA ensured that space on panels, and
control over the content of those panels, would remain within the section's hands. This
largely removed the complaint that there were no conference outlets for methodological
Page 17 of 21

work (especially with the Summer Meetings in addition to APSA). The organized section
was also able to give awards for outstanding work in the field, an important element of
professional recognition.
These organizational developments have promoted the development of political

methodology for over twenty years now. In that time, the field has grown considerably
(p. 812) judged by number of practitioners, number of articles, and by attendance at the
Summer Methodology Meetings. On the more critical intellectual front, the field has
produced a substantial number of advances since 1984. Among these are new models for
ecological inference, scaling of roll‐call votes, sample selection models, Bayesian
methods, and causal inference to name only a few.
The existence of the Society for Political Methodology has also increased expectations for
graduate training, at least among those who see their careers as methodologists. While
lack of strong fundamental skills remains a problem at the ICPSR Summer Program,
those who seek professional success within methodology are now much better trained as
graduate students than was the case twenty years ago. At the first Summer Meeting that
welcomed them, graduate students concluded “we have to learn more math.” That
experience has continued with each successive cohort attending the Summer Meetings,
and it has helped raise the ante for political methodology.
Within the profession, political methodology now holds a secure place, if not yet one that
is universally welcomed. By its institutional role through the APSA Organized Section,
political methodology both literally and figuratively has a seat at the table when
conferences are planned, officers elected, and awards handed out. These are critical
elements of professional presence and recognition, and they did not exist in 1983. More
recently methodology has been added to departmental reputational rankings as a field
alongside comparative, IR, theory, and American. This reflects the increased appreciation
of the role of methodology within the field, and the extent to which a graduate
department is hobbled if it lacks strength in quantitative methods training and research.
Despite these advances, political methodology continues to suffer from the rarity of
strong quantitative training at the undergraduate level. Only a handful of students enter
graduate school with more than a single semester of calculus or more than one
introduction to statistics course. The vast majority of graduate programs spend the first
year of methodology training on topics that should be well within the reach of
undergraduates, had they only taken the classes. And because few students have strong
mathematical training, the amount of time that must be spent on remedial training in
math (often in an ad hoc and self‐study manner) is substantial and delays or limits the
level to which students can rise in statistical training. In this, political science continues
to compare unfavorably with economics, in which substantial mathematical and statistical
education occurs at the undergraduate level.
Page 18 of 21

These limitations in undergraduate training will continue to hamper the development of

political methodology. Excellent students must work harder than they should to catch up
once in graduate school. In this area, political methodology may yet not have had the
impact on the discipline that it needs to have in order to continue to raise the level of
methodological sophistication. It is not necessary that all undergraduate majors in
political science take advanced mathematics or statistics. But it is highly desirable that
those who might consider graduate careers know that their options will be considerably
enhanced by a more sophisticated undergraduate education.
References
ACHEN, C. H. 1981. Toward theories of data. In Political Science: The State of the
Discipline, ed. A. W. Finifter. Washington, DC: American Political Science Association.
—— 1983. If party ID influences the vote, Goodman's ecological regression is biased (but
factor analysis is consistent). Presented at the Annual Meeting of the American Political
Science Association, Chicago.
AUSTIN, E. W. 2001. From Hollerith to the Web: a half‐century of electronic social

science data archives. Presented at the Sawyer Seminar on Archives, Documentation and
the Institutions of Social Memory, University of Michigan, February 14.
BARTELS, L. M., and BRADY, H. E. 1993. The state of quantitative political

methodology. In Political Science: The State of the Discipline II, ed. A. W. Finifter.
Washington, DC: American Political Science Association.
BLALOCK, H. M. 1989. Report of the ICPSR Review Committee. ICPSR, June.
BRADY, H. E., and COLLIER, D. (eds.) 2004. Rethinking Social Inquiry: Diverse Tools,
Shared Standards. Lanham, Md.: Rowman and Littlefield.
CAMPBELL, A., and KAHN, R. L. 1952. The People Elect a President. Ann Arbor, Mich.:
Institute for Social Research.
—— GURIN, G., and MILLER, W. E. 1954. The Voter Decides. Evanston, Ill.: Row,
Peterson.
—— CONVERSE, P. E., MILLER, W. E., and STOKES, D. E. 1960. The American Voter.
New York: John Wiley.
DAHL, R. A. 1961. Who Governs? New Haven, Conn.: Yale University Press.
FINIFTER, A. W. (ed.) 1981. Political Science: The State of the Discipline. Washington,
DC: American Political Science Association.
GOSNELL, H. F. 1927. Getting out the Vote: An Experiment in the Stimulation of Voting.
Chicago: University of Chicago Press.
Page 19 of 21

HANUSHEK, E. A., and JACKSON, J. E. 1977. Statistical Methods for Social Scientists.
New York: Academic Press.
KEY, V. O. 1949. Southern Politics. New York: Alfred A. Knopf.
—— 1954. A Primer of Statistics for Political Scientists. New York: Thomas Y. Crowell.
—— 1961. Public Opinion and American Democracy. New York: Alfred A. Knopf.
KING, G. 1990. On political methodology. Political Analysis, 2: 1–30.
—— KEOHANE, R. O., and VERBA, S. 1994. Designing Social Inquiry: Scientific

Inference in Qualitative Research. Princeton, NJ: Princeton University Press.
LAZARSFELD, P., BERELSON, B., and GAUDET, H. 1944. The People's Choice. New
York: Columbia University Press.
RICE, S. A. 1928. Quantitative Methods in Politics. New York: A. A. Knopf.
Notes:
(1) The Society for Political Methodology has the same officers as the organized section.
It exists as a separate entity in support of the Summer Methodology Meetings and
maintains an extensive working paper archive, newsletter, and website. For our purposes
here the Society and the organized section are the same. As a historical matter, the
Society existed as the organizer of the Summer Methodology Meetings prior to the
formation of the Organized Section.
(2) See Key's (1961) preface, pp. viii– ix, for both a description of his access and his
acknowledgement of appreciation.
(3) Some of the variation in this trend is undoubtedly due to the changing mix of classes
sponsored by grants, such as programs in criminal justice, which targeted faculty or
Ph.D.s. The basic fact remains that as many as a quarter of participants are holders of the
Ph.D.
(4) This of course means that easy inferences about how political science has changed
based on Summer Program participation faces an ecological inference fallacy.
(5) Henry Heitowit comments, “Now it can be told: Reminiscences on 80 years at ICPSR,”
Erik Austin and Henry Heitowit, Panel Presentation at the 2007 ICPSR Meeting of Official
Representatives, Ann Arbor, October 18–20, 2007.
Charles H. Franklin
Page 20 of 21

Charles H. Franklin is Professor of Political Science, University of Wisconsin,

Madison.
Page 21 of 21

Oxford Handbooks Online: Quantitative Methodology

Uploaded by

Copyright:

Available Formats

Oxford Handbooks Online: Quantitative Methodology

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Oxford Handbooks Online: Quantitative Methodology

Uploaded by

Copyright:

Available Formats

Quantitative Methodology

Oxford Handbooks Online

Print Publication Date: Aug 2008

Abstract and Keywords

Subscriber: Vanderbilt University Library; date: 17 January 2019

developments in quantitative methodology proceeded largely apart from developments in

Subscriber: Vanderbilt University Library; date: 17 January 2019

Subscriber: Vanderbilt University Library; date: 17 January 2019

2 The Development of Summer Programs in

Subscriber: Vanderbilt University Library; date: 17 January 2019

approach to political science, and electoral behavior, without giving it an exaggerated

In the development of quantitative training programs, however, Michigan and the

Subscriber: Vanderbilt University Library; date: 17 January 2019

As in all successful collaborations there was to be a two‐way flow of benefits

Subscriber: Vanderbilt University Library; date: 17 January 2019

The second core

Subscriber: Vanderbilt University Library; date: 17 January 2019

If the impetus for the

Subscriber: Vanderbilt University Library; date: 17 January 2019

Figure 35.2 shows the

(p. 804) 2.1 Dynamics of Course Offerings

Subscriber: Vanderbilt University Library; date: 17 January 2019

Subscriber: Vanderbilt University Library; date: 17 January 2019

2.2 Why do we Continue to Need Summer Programs?

Subscriber: Vanderbilt University Library; date: 17 January 2019

While only a small minority of undergraduates go on to graduate school in the social

If something like a quarter of Summer Program students require introductory classes,

Subscriber: Vanderbilt University Library; date: 17 January 2019

For well‐prepared students, summer programs offer a relatively quick introduction to

Subscriber: Vanderbilt University Library; date: 17 January 2019

Some degree of “remedial” instruction in summer programs is likely to remain. This

3 The Society for Political Methodology and

Subscriber: Vanderbilt University Library; date: 17 January 2019

Subscriber: Vanderbilt University Library; date: 17 January 2019

Subscriber: Vanderbilt University Library; date: 17 January 2019

Subscriber: Vanderbilt University Library; date: 17 January 2019

These organizational developments have promoted the development of political

Subscriber: Vanderbilt University Library; date: 17 January 2019

These limitations in undergraduate training will continue to hamper the development of

AUSTIN, E. W. 2001. From Hollerith to the Web: a half‐century of electronic social

BARTELS, L. M., and BRADY, H. E. 1993. The state of quantitative political

BLALOCK, H. M. 1989. Report of the ICPSR Review Committee. ICPSR, June.

Subscriber: Vanderbilt University Library; date: 17 January 2019

KEY, V. O. 1949. Southern Politics. New York: Alfred A. Knopf.

KING, G. 1990. On political methodology. Political Analysis, 2: 1–30.

—— KEOHANE, R. O., and VERBA, S. 1994. Designing Social Inquiry: Scientific

RICE, S. A. 1928. Quantitative Methods in Politics. New York: A. A. Knopf.

Subscriber: Vanderbilt University Library; date: 17 January 2019

Charles H. Franklin is Professor of Political Science, University of Wisconsin,

Subscriber: Vanderbilt University Library; date: 17 January 2019

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.